Broadcasting compatibility check in NumPy - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is broadcasting in numpy?

Broadcasting is a way numpy allows arrays of different shapes to work together in arithmetic operations by automatically expanding their shapes to be compatible.

intermediate

How does numpy check if two arrays are broadcast compatible?

Numpy compares their shapes from right to left. For each dimension, the sizes must be equal or one of them must be 1. If this holds for all dimensions, arrays are broadcast compatible.

beginner

Example: Are arrays with shapes (3, 1) and (1, 4) broadcast compatible?

Yes. Comparing from right: 1 and 4 (one is 1), 3 and 1 (one is 1). So they can broadcast to shape (3, 4).

beginner

What happens if arrays are not broadcast compatible?

Numpy raises a ValueError indicating operands could not be broadcast together because their shapes are incompatible.

intermediate

How can you programmatically check broadcasting compatibility in numpy?

You can try to use numpy.broadcast_shapes(shape1, shape2) which returns the broadcasted shape if compatible, or raises an error if not.

Which condition must be true for each dimension when checking broadcasting compatibility?

A Dimensions are both prime numbers

B Dimensions differ by exactly 1

C Dimensions are both even numbers

D Dimensions are equal or one is 1

Broadcasting requires each dimension to be equal or one of them to be 1.

Are arrays with shapes (5, 3) and (3,) broadcast compatible?

A Only if 5 equals 3

B No

C Yes

D Only if both have same number of dimensions

Shape (3,) is treated as (1,3), so dimensions are compatible.

What error does numpy raise if arrays are not broadcast compatible?

A ValueError

B TypeError

C IndexError

D KeyError

Numpy raises ValueError for incompatible broadcasting.

What is the broadcasted shape of arrays with shapes (4,1,6) and (3,1)?

A Not compatible

B (4,1,6)

C (3,1,6)

D (4,3,6)

Arrays with shapes (4,1,6) and (3,1) are not broadcast compatible because after padding (3,1) to (1,3,1), the second dimension comparison is 1 vs 3 (compatible), but the first dimension comparison is 4 vs 1 (compatible), but the third dimension comparison is 6 vs 1 (compatible). However, the second dimension 1 vs 3 is compatible, but the first dimension 4 vs 1 is compatible, so the broadcasted shape would be (4,3,6). Actually, the original explanation was incorrect because (3,1) padded to (1,3,1) compared to (4,1,6) yields dimensions: 4 vs 1 (compatible), 1 vs 3 (compatible), 6 vs 1 (compatible), so they are compatible and broadcasted shape is (4,3,6). Therefore, the correct answer is A.

Which numpy function can be used to check broadcast compatibility of shapes?

A numpy.check_broadcast

B numpy.broadcast_shapes

C numpy.shape_compat

D numpy.broadcast_check

numpy.broadcast_shapes returns the broadcasted shape or raises an error if incompatible.

Explain how numpy determines if two arrays can be broadcast together.

Think about matching dimensions starting from the end.

Describe a real-life example where broadcasting helps in data science calculations.

Consider adding a list of values to multiple data rows.